Lie Theory

Lie theory is the mathematical study of continuous symmetry — the symmetry of smooth objects under continuous transformations. Born in the 1870s from Sophus Lie’s audacious attempt to apply Galois theory to differential equations, it has grown into one of the deepest and most far-reaching structures in all of mathematics, touching differential geometry, number theory, quantum physics, and the very foundations of modern representation theory. The central insight is deceptively simple: a smooth group can be studied locally, near the identity element, through its linear approximation — a Lie algebra — and this linearization translates hard nonlinear problems into linear algebra.

Lie Groups and Their Properties

A Lie group is simultaneously a smooth manifold and a group, with the requirement that the group operations — multiplication and inversion — are smooth maps. This blending of algebra and geometry is what makes the theory so powerful. The smoothness condition is not merely a technical nicety; it is what allows calculus to enter group theory, enabling the definition of tangent vectors, differential forms, and one-parameter subgroups.

The canonical examples are the matrix Lie groups — closed subgroups of $GL(n, \mathbb{R})$ or $GL(n, \mathbb{C})$, the general linear group of invertible $n \times n$ matrices. The special linear group $SL(n, \mathbb{F}) = \{A \in GL(n, \mathbb{F}) : \det(A) = 1\}$ is an $(n^2 - 1)$-dimensional Lie group. The orthogonal group $O(n) = \{A \in GL(n, \mathbb{R}) : A^T A = I\}$ and its connected component $SO(n)$ encode the rotational symmetries of Euclidean space. The unitary group $U(n) = \{A \in GL(n, \mathbb{C}) : A^\dagger A = I\}$ and the special unitary group $SU(n)$ are their complex counterparts. The symplectic group $Sp(2n, \mathbb{F})$ preserves a non-degenerate skew-symmetric bilinear form and plays a central role in Hamiltonian mechanics. These classical groups, together with the five exceptional Lie groups, form the complete landscape of simple Lie groups.

A fundamental result is the closed subgroup theorem (also called the Cartan theorem after Élie Cartan, who proved it in 1930): any closed subgroup of a Lie group is itself a Lie group. This theorem means one can specify groups by algebraic conditions — such as the determinant being 1, or a matrix being orthogonal — and be guaranteed a smooth manifold structure for free.

Topological properties of Lie groups are essential context. A Lie group may be compact (like $SO(n)$, $SU(n)$, and $Sp(n)$) or non-compact (like $SL(n, \mathbb{R})$ and $GL(n, \mathbb{R})$). Compact Lie groups are analytically better-behaved: every representation is completely reducible, and the group carries a unique probability measure invariant under left translation (the Haar measure). Every connected Lie group has a universal covering group, which is a simply-connected Lie group with the same local structure. The classical example is $SU(2)$ covering $SO(3)$ with a degree-two covering map — a fact that explains the mysterious half-integer spin of electrons in quantum mechanics.

Lie Algebras and Structure Theory

The Lie algebra $\mathfrak{g}$ of a Lie group $G$ is the tangent space at the identity element, equipped with an additional operation called the Lie bracket. Concretely, for a matrix group, $\mathfrak{g}$ consists of matrices $X$ such that $e^{tX} \in G$ for all real $t$, and the bracket is simply the matrix commutator $[X, Y] = XY - YX$. More abstractly, a Lie algebra is a vector space equipped with a bracket $[\cdot, \cdot]: \mathfrak{g} \times \mathfrak{g} \to \mathfrak{g}$ satisfying three axioms: bilinearity, antisymmetry $[X, Y] = -[Y, X]$, and the Jacobi identity

$[X, [Y, Z]] + [Y, [Z, X]] + [Z, [X, Y]] = 0.$

The Jacobi identity is the infinitesimal echo of associativity in the group, and it is precisely what distinguishes a Lie algebra from a mere anti-commutative algebra.

The passage from groups to algebras is made concrete by the exponential map $\exp: \mathfrak{g} \to G$, defined by the matrix exponential $\exp(X) = \sum_{k=0}^\infty X^k / k!$ for matrix groups, and more generally by following the flow of the left-invariant vector field generated by $X$ for unit time. Near the identity, the exponential map is a diffeomorphism, meaning $\mathfrak{g}$ captures the full local structure of $G$. The relationship between two group elements near the identity is governed by the Baker-Campbell-Hausdorff formula:

$\log(\exp(X)\exp(Y)) = X + Y + \tfrac{1}{2}[X,Y] + \tfrac{1}{12}([X,[X,Y]] + [Y,[Y,X]]) + \cdots,$

an infinite series whose terms involve only nested Lie brackets. This formula shows that the group multiplication, locally, is entirely determined by the Lie algebra structure.

The intrinsic structure of a Lie algebra is studied through its derived series and lower central series. A Lie algebra $\mathfrak{g}$ is solvable if its derived series $\mathfrak{g} \supset [\mathfrak{g},\mathfrak{g}] \supset [[\mathfrak{g},\mathfrak{g}],[\mathfrak{g},\mathfrak{g}]] \supset \cdots$ terminates at zero; it is nilpotent if its lower central series terminates at zero. Engel’s theorem (1890) characterizes nilpotent algebras: $\mathfrak{g}$ is nilpotent if and only if every element acts nilpotently (i.e., as a nilpotent endomorphism) under the adjoint representation $\mathrm{ad}_X(Y) = [X, Y]$.

The radical of $\mathfrak{g}$, denoted $\mathrm{rad}(\mathfrak{g})$, is the largest solvable ideal. A Lie algebra is semisimple if its radical is zero. The Levi decomposition theorem (proved by Eugenio Levi in 1905 and corrected by Lie) states that any finite-dimensional Lie algebra splits as $\mathfrak{g} = \mathrm{rad}(\mathfrak{g}) \rtimes \mathfrak{s}$, a semidirect product of the radical with a semisimple complement $\mathfrak{s}$. This reduces the classification problem to understanding solvable algebras and semisimple algebras separately — and it is the semisimple case that yields one of the most beautiful classification theorems in mathematics.

Root Systems and Cartan Classification

The classification of simple Lie algebras over $\mathbb{C}$ is one of the crowning achievements of 19th and early 20th century mathematics, completed by Killing and Cartan between 1888 and 1894. The key tool is the combinatorial data encoded in a root system.

Given a complex semisimple Lie algebra $\mathfrak{g}$, a Cartan subalgebra $\mathfrak{h}$ is a maximal abelian subalgebra consisting entirely of semisimple (diagonalizable) elements. All Cartan subalgebras of a semisimple algebra are conjugate under inner automorphisms, so the dimension of $\mathfrak{h}$, called the rank of $\mathfrak{g}$, is well-defined. The Cartan subalgebra acts on $\mathfrak{g}$ by the adjoint representation, and since elements of $\mathfrak{h}$ are simultaneously diagonalizable, $\mathfrak{g}$ decomposes into a direct sum of simultaneous eigenspaces:

$\mathfrak{g} = \mathfrak{h} \oplus \bigoplus_{\alpha \in \Phi} \mathfrak{g}_\alpha,$

where $\Phi \subset \mathfrak{h}^*$ is the finite set of roots — the nonzero linear functionals $\alpha$ for which $\mathfrak{g}_\alpha = \{X \in \mathfrak{g} : [H, X] = \alpha(H) X \text{ for all } H \in \mathfrak{h}\}$ is nonzero. Each root space $\mathfrak{g}_\alpha$ turns out to be one-dimensional, and the algebra is spanned by these root spaces together with $\mathfrak{h}$.

An abstract root system is a finite subset $\Phi$ of a Euclidean space $V$ satisfying four axioms: the roots span $V$; if $\alpha \in \Phi$, then $-\alpha \in \Phi$ and the only scalar multiples of $\alpha$ in $\Phi$ are $\pm \alpha$; every root $\alpha$ defines a reflection $s_\alpha$ that preserves $\Phi$; and for all $\alpha, \beta \in \Phi$, the Cartan integer $\langle \beta, \alpha \rangle = 2(\beta, \alpha)/(\alpha, \alpha)$ is an integer. This last condition, the crystallographic restriction, severely constrains the possible angles between roots: the angle between two roots must be a multiple of $30°$.

A set of simple roots $\Delta \subset \Phi$ is a basis for $V$ such that every root is either a non-negative or a non-positive integer linear combination of elements of $\Delta$. The Cartan integers between simple roots are encoded in the Cartan matrix $A$, with entries $A_{ij} = \langle \alpha_i, \alpha_j \rangle$, and this matrix is displayed visually as a Dynkin diagram: one node per simple root, with $A_{ij} A_{ji}$ edges between nodes $i$ and $j$ (0, 1, 2, or 3), and an arrow pointing toward the shorter root when roots have different lengths.

The classification of irreducible root systems — equivalently, of complex simple Lie algebras — is then a finite combinatorial problem. The complete list consists of four infinite families and five exceptional cases:

• $A_n$ ($n \geq 1$): corresponds to $\mathfrak{sl}(n+1)$, with Dynkin diagram a straight chain of $n$ nodes.
• $B_n$ ($n \geq 2$): corresponds to $\mathfrak{so}(2n+1)$, with a chain ending in a double edge with arrow.
• $C_n$ ($n \geq 3$): corresponds to $\mathfrak{sp}(2n)$, with a chain ending in a double edge with arrow in the other direction.
• $D_n$ ($n \geq 4$): corresponds to $\mathfrak{so}(2n)$, with a chain ending in a fork.
• Exceptionals: $G_2$, $F_4$, $E_6$, $E_7$, $E_8$, discovered by Wilhelm Killing in 1888 and constructed explicitly by Cartan.

The Weyl group $W$ of a root system is the group generated by the reflections $s_\alpha$ for all $\alpha \in \Phi$. It is a finite group acting faithfully on the root system, and it encodes the symmetry of the Dynkin diagram. For $A_n$, the Weyl group is the symmetric group $S_{n+1}$; for $B_n$ and $C_n$, it is the hyperoctahedral group.

Representation Theory of Lie Groups and Algebras

The theory of representations — how Lie groups and algebras act as linear transformations on vector spaces — is where the abstract theory meets its most spectacular applications.

A representation of a Lie algebra $\mathfrak{g}$ on a vector space $V$ is a Lie algebra homomorphism $\rho: \mathfrak{g} \to \mathfrak{gl}(V)$, where $\mathfrak{gl}(V)$ denotes the space of all linear endomorphisms of $V$ with the commutator bracket. The fundamental organizing principle for finite-dimensional representations of semisimple algebras is the theory of highest weights, developed by Élie Cartan in the early 1900s. For a fixed Cartan subalgebra $\mathfrak{h}$ and a choice of positive roots, every element of $\mathfrak{h}$ acts diagonally on any finite-dimensional representation $V$, decomposing it into weight spaces:

$V = \bigoplus_{\mu \in \mathfrak{h}^*} V_\mu, \quad V_\mu = \{v \in V : H \cdot v = \mu(H) v \text{ for all } H \in \mathfrak{h}\}.$

The dominant integral weights $\lambda$ satisfying $\langle \lambda, \alpha^\vee \rangle \in \mathbb{Z}_{\geq 0}$ for all positive coroots $\alpha^\vee$ parametrize the irreducible representations. The highest weight theorem states that every irreducible finite-dimensional representation $L(\lambda)$ is uniquely determined by its highest weight $\lambda$, and conversely, every dominant integral weight arises as the highest weight of exactly one irreducible. The highest weight vector $v_\lambda \in L(\lambda)$ is annihilated by all positive root operators and generates the entire representation by the action of negative root operators.

The Weyl character formula, proved by Hermann Weyl in 1925, gives the character — the function $\chi_\lambda(H) = \mathrm{tr}_{L(\lambda)}(e^H)$ — of an irreducible representation explicitly:

$\chi_\lambda = \frac{\sum_{w \in W} (-1)^{\ell(w)} e^{w(\lambda + \rho) - \rho}}{\prod_{\alpha > 0}(1 - e^{-\alpha})},$

where $\rho = \frac{1}{2}\sum_{\alpha > 0} \alpha$ is the Weyl vector (the half-sum of positive roots) and $\ell(w)$ is the length of the Weyl group element $w$. From this formula, the Weyl dimension formula yields the dimension of $L(\lambda)$ as a product over positive roots:

$\dim L(\lambda) = \prod_{\alpha > 0} \frac{\langle \lambda + \rho, \alpha \rangle}{\langle \rho, \alpha \rangle}.$

For compact Lie groups, the Peter-Weyl theorem (proved by Fritz Peter and Hermann Weyl in 1927) plays an analogous role in infinite-dimensional harmonic analysis: the matrix coefficients of irreducible unitary representations form an orthonormal basis for $L^2(G)$ with respect to the Haar measure. This theorem is the non-abelian generalization of Fourier analysis, reducing harmonic analysis on a compact group to the algebraic classification of its irreducible representations.

Classical Lie Groups and Algebras

The classical Lie groups — $SL(n)$, $SO(n)$, $SU(n)$, and $Sp(2n)$ — deserve individual study both as the prototypical examples of the theory and for their direct role in geometry and physics.

The special linear group $SL(2, \mathbb{C})$ is the simplest non-abelian semisimple Lie group and the testing ground for the entire theory. Its Lie algebra $\mathfrak{sl}(2, \mathbb{C})$ is spanned by three elements

$e = \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}, \quad f = \begin{pmatrix} 0 & 0 \\ 1 & 0 \end{pmatrix}, \quad h = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix},$

satisfying $[h, e] = 2e$, $[h, f] = -2f$, $[e, f] = h$. The irreducible representations are labeled by non-negative integers $n$: the unique $(n+1)$-dimensional representation $V_n$ has a basis of weight vectors with weights $n, n-2, \ldots, -n$. The half-integer-weight representations of $\mathfrak{sl}(2)$ lift to representations of the universal cover $\widetilde{SL}(2, \mathbb{R}) \cong SU(2)$ but not of $SO(3)$ itself — this is the algebraic explanation for spin-$\frac{1}{2}$ particles in physics.

The orthogonal and spin groups are linked by a double covering. For $n \geq 3$, the spin group $\mathrm{Spin}(n)$ is the universal double cover of $SO(n)$: there is a surjective homomorphism $\mathrm{Spin}(n) \to SO(n)$ with kernel $\{+1, -1\} \cong \mathbb{Z}/2\mathbb{Z}$. The spinor representation of $\mathrm{Spin}(n)$ does not descend to a representation of $SO(n)$ — its weights are half-integers — and this is responsible for the existence of fermionic fields in physics. In low dimensions there are exceptional isomorphisms: $\mathrm{Spin}(3) \cong SU(2)$, $\mathrm{Spin}(4) \cong SU(2) \times SU(2)$, $\mathrm{Spin}(5) \cong Sp(4)$, and $\mathrm{Spin}(6) \cong SU(4)$, reflecting the coincidences $B_1 \cong C_1$, $D_2 \cong A_1 \times A_1$, $B_2 \cong C_2$, and $D_3 \cong A_3$ in the Dynkin diagram classification.

The unitary groups play a central role in quantum mechanics. The group $SU(2)$ is diffeomorphic to the 3-sphere $S^3$ and acts as the group of unit quaternions. The irreducible representations of $SU(n)$ are parametrized by Young diagrams — combinatorial objects encoding the partitions of non-negative integers — and their characters are given by the Schur polynomials, connecting representation theory to algebraic combinatorics. The Clebsch-Gordan decomposition of tensor products into irreducibles, and the branching rules for restricting representations to subgroups, are of direct computational importance in quantum chemistry and particle physics.

Symmetric Spaces and Geometry

The study of symmetric spaces, initiated by Élie Cartan in his series of papers from 1926 to 1935, reveals how Lie theory governs the global geometry of the most important Riemannian manifolds.

A Riemannian symmetric space is a Riemannian manifold $M$ such that at every point $p \in M$ there exists an isometry $s_p: M \to M$ that fixes $p$ and reverses all geodesics through $p$ — that is, $ds_p|_p = -\mathrm{Id}$. The key consequence is that such a space is homogeneous: its isometry group $G$ acts transitively, so $M \cong G/K$ for $K$ the stabilizer of any fixed point. The involution $\theta: G \to G$ defined by $\theta(g) = s_p \circ g \circ s_p$ is a Lie group automorphism with $\theta^2 = \mathrm{Id}$, and it induces a decomposition of the Lie algebra $\mathfrak{g} = \mathfrak{k} \oplus \mathfrak{p}$, where $\mathfrak{k}$ is the $+1$-eigenspace (the Lie algebra of $K$) and $\mathfrak{p}$ is the $-1$-eigenspace (the tangent space at the base point). The conditions $[\mathfrak{k}, \mathfrak{k}] \subset \mathfrak{k}$, $[\mathfrak{k}, \mathfrak{p}] \subset \mathfrak{p}$, and $[\mathfrak{p}, \mathfrak{p}] \subset \mathfrak{k}$ encode the entire geometric structure of the symmetric space in purely algebraic terms — this is the Cartan decomposition.

Symmetric spaces come in two flavors. A compact type symmetric space (such as the sphere $S^n \cong SO(n+1)/SO(n)$, the Grassmannian $\mathrm{Gr}(k,n) \cong O(n)/(O(k) \times O(n-k))$, and the complex projective space $\mathbb{CP}^n \cong SU(n+1)/U(n)$) has non-negative sectional curvature. Its non-compact dual (such as real hyperbolic space $\mathbb{H}^n \cong SO(n,1)/SO(n)$, the Siegel upper half-space, and the space of positive definite matrices) has non-positive sectional curvature. The duality between compact and non-compact symmetric spaces — swapping $\mathfrak{p} \mapsto i\mathfrak{p}$ in the Cartan decomposition — is a fundamental symmetry of the theory, analogous to the relationship between the sphere and hyperbolic space.

Cartan’s complete classification of symmetric spaces yielded a finite list: the classical spaces associated with the four families of classical groups, plus twelve exceptional spaces associated with the exceptional Lie groups, including the octonion projective plane $\mathbb{OP}^2 \cong F_4/\mathrm{Spin}(9)$. The geometry of geodesics on symmetric spaces is governed entirely by the root system: geodesics through the base point are parametrized by elements of $\mathfrak{p}$, and the root system of the symmetric space (which may differ from the root system of $G$) controls the flat submanifolds (the flats or maximal tori) and the curvature tensor explicitly.

Algebraic Groups

The theory of algebraic groups transports Lie theory into algebraic geometry, replacing smooth manifolds by algebraic varieties and allowing the base field to be arbitrary — in particular, finite fields or the $p$-adic numbers.

An algebraic group $G$ over a field $k$ is an algebraic variety over $k$ equipped with morphisms $\mu: G \times G \to G$ (multiplication) and $\iota: G \to G$ (inversion) that are regular maps of varieties and satisfy the group axioms. The classical matrix groups — $GL_n$, $SL_n$, $SO_n$, $Sp_{2n}$ — are all algebraic groups, defined by polynomial equations on the space of matrices.

The structure theory of reductive algebraic groups over algebraically closed fields, developed primarily by Claude Chevalley in the 1950s and Armand Borel and Jacques Tits in the 1960s, mirrors the Lie algebra classification remarkably faithfully. The central notion is a Borel subgroup $B \subset G$ — a maximal solvable connected algebraic subgroup. The quotient variety $G/B$ is a projective variety called the flag variety (since for $G = GL_n$, it parametrizes complete flags $0 = V_0 \subset V_1 \subset \cdots \subset V_n = k^n$). The Bruhat decomposition $G = \bigsqcup_{w \in W} BwB$ decomposes $G$ into double cosets indexed by the Weyl group, and the corresponding decomposition of the flag variety $G/B = \bigsqcup_{w \in W} Bw \cdot [B]$ into Schubert cells provides a cell decomposition of this projective variety.

Over finite fields $\mathbb{F}_q$, algebraic groups give rise to finite groups of Lie type — the groups $GL_n(\mathbb{F}_q)$, $SL_n(\mathbb{F}_q)$, $SO_n(\mathbb{F}_q)$, and so on — which account for the vast majority of finite simple groups in the classification of finite simple groups. Chevalley’s construction of these groups over arbitrary fields unified the theory and produced several new families (the Chevalley groups and Steinberg variations). The representation theory of algebraic groups — particularly in characteristic $p > 0$, where complete reducibility can fail — remains an active area of research, with Lusztig’s conjecture on modular representations only partially resolved.

The algebraic group perspective also leads naturally to the Langlands program, the vast web of conjectures (and theorems) connecting number theory, automorphic forms, and representation theory of reductive groups over local and global fields — arguably the deepest open problem in modern mathematics.